Merge pull request #290 from FJShen/289-tracer-missing-kernels-in-lat…

…est-dev-on-ubuntu-1804

Fixes issue 289: tracer missing kernels in latest dev on ubuntu 1804

.github/workflows/long-tests.yml

@@ -29,22 +29,22 @@ jobs:
           git checkout cluster-ubuntu
       - name: Build Accel-Sim
         run: |
-          source ./env-setup/11.2.1_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           rm -rf ./gpu-simulator/gpgpu-sim
           source ./gpu-simulator/setup_environment.sh
           make clean -C gpu-simulator
           srun -c20 make -j20 -C gpu-simulator
       - name: Run SASS
         run: |
-          source ./env-setup/11.2.1_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           source ./gpu-simulator/setup_environment.sh
           ./util/job_launching/run_simulations.py -B rodinia_2.0-ft,GPU_Microbenchmark -C QV100-SASS -T ~/../common/accel-sim/traces/volta-tesla-v100/latest/ -N sass-short-$GITHUB_RUN_NUMBER"_"$GITHUB_RUN_ATTEMPT
           ./util/job_launching/run_simulations.py -B rodinia_2.0-ft,GPU_Microbenchmark -C RTX2060-SASS -T ~/../common/accel-sim/traces/turing-rtx2060/latest/ -N sass-short-$GITHUB_RUN_NUMBER"_"$GITHUB_RUN_ATTEMPT
           ./util/job_launching/run_simulations.py -B rodinia_2.0-ft,GPU_Microbenchmark -C RTX3070-SASS -T ~/../common/accel-sim/traces/ampere-rtx3070/latest/ -N sass-short-$GITHUB_RUN_NUMBER"_"$GITHUB_RUN_ATTEMPT
           ./util/job_launching/monitor_func_test.py -v -s stats-per-app-sass.csv -N sass-short-$GITHUB_RUN_NUMBER"_"$GITHUB_RUN_ATTEMPT
       - name: Archive Stats
         run: |
-          source ./env-setup/11.2.1_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           rm -rf ./statistics-archive
           git clone [email protected]:accel-sim/statistics-archive.git
           # either create a new branch or check it out if it already exists
@@ -65,7 +65,7 @@ jobs:
           git -C ./statistics-archive push -u origin git_${GITHUB_REF}"_"$GITHUB_RUN_NUMBER"_"$GITHUB_RUN_ATTEMPT
       - name: Correlate Ubench
         run: |
-          source ./env-setup/11.2.1_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           ./util/hw_stats/get_hw_data.sh > /dev/null
           rm -rf ./util/plotting/correl-html/
           ./util/plotting/plot-correlation.py -c ./statistics-archive/ubench/v100-ubench-sass.csv -H ./hw_run/QUADRO-V100/device-0/10.2/ | tee v100-ubench-correl.txt
@@ -90,14 +90,14 @@ jobs:
           git checkout cluster-ubuntu
       - name: Build Accel-Sim
         run: |
-          source ./env-setup/11.2.1_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           rm -rf ./gpu-simulator/gpgpu-sim
           source ./gpu-simulator/setup_environment.sh
           make clean -C gpu-simulator
           srun -c20 make -j20 -C gpu-simulator
       - name: Run PTX
         run: |
-          source ./env-setup/11.2.1_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           source ./gpu-simulator/setup_environment.sh
 
           rm -rf ./gpu-app-collection

.github/workflows/tracer-tool.yml

@@ -29,14 +29,14 @@ jobs:
           git checkout cluster-ubuntu
       - name: Build Accel-Sim
         run: |
-          source ./env-setup/11.0_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           rm -rf ./gpu-simulator/gpgpu-sim
           source ./gpu-simulator/setup_environment.sh
           make clean -C gpu-simulator
           make -j -C gpu-simulator
       - name: test-prebuilt-traces
         run: |
-          source ./env-setup/11.0_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           source ./gpu-simulator/setup_environment.sh
           ./get-accel-sim-traces.py -a tesla-v100/rodinia_2.0-ft
           cd hw_run; tar -xzvf rodinia_2.0-ft.tgz; cd -
@@ -46,32 +46,32 @@ jobs:
           rm -rf sim_run_11.0
       - name: Build Tracer
         run: |
-          source ./env-setup/11.0_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           ./util/tracer_nvbit/install_nvbit.sh
           make clean -C ./util/tracer_nvbit/
           make -C ./util/tracer_nvbit/
       - name: rodinia_2.0-ft-build
         run: |
-          source ./env-setup/11.0_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           rm -rf ./gpu-app-collection/
           git clone [email protected]:accel-sim/gpu-app-collection.git
           source ./gpu-app-collection/src/setup_environment
           ln -s /home/tgrogers-raid/a/common/data_dirs ./gpu-app-collection/
           make -C ./gpu-app-collection/src rodinia_2.0-ft
       - name: generate-rodinia_2.0-ft-traces
         run: |
-          source ./env-setup/11.0_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           source ./gpu-app-collection/src/setup_environment
           rm -rf ./hw_run/
           ./util/tracer_nvbit/run_hw_trace.py -B rodinia_2.0-ft -D 7
       - name: generate-rodinia_2.0-ft-hw_stats
         run: |
-          source ./env-setup/11.0_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           source ./gpu-app-collection/src/setup_environment
           ./util/hw_stats/run_hw.py -B rodinia_2.0-ft -D 7
       - name: test-new-traces
         run: |
-          source ./env-setup/11.0_env_setup.sh
+          source ./env-setup/11.7.0_env_setup.sh
           source ./gpu-simulator/setup_environment.sh
-          ./util/job_launching/run_simulations.py -B rodinia_2.0-ft -C QV100-SASS -T ./hw_run/traces/device-7/11.0/ -N rodinia_2.0-ft-$$
+          ./util/job_launching/run_simulations.py -B rodinia_2.0-ft -C QV100-SASS -T ./hw_run/traces/device-7/ -N rodinia_2.0-ft-$$
           ./util/job_launching/monitor_func_test.py -I -v -s rodinia-stats-per-app.csv -N rodinia_2.0-ft-$$

gpu-simulator/trace-parser/trace_parser.cc

@@ -343,6 +343,7 @@ kernel_trace_t *trace_parser::parse_kernel_info(
   // Important to clear the istream. Otherwise, the eofbit from the last 
   // kernel may be carried over to this kernel
   ifs->clear();
+  clearerr(stdin);
   while (!ifs->eof()) {
     getline(*ifs, line);
 

util/tracer_nvbit/tracer_tool/Makefile

@@ -17,7 +17,7 @@ NVCC_PATH=-L $(subst bin/nvcc,lib64,$(shell which nvcc | tr -s /))
 SOURCES=$(wildcard *.cu)
 
 OBJECTS=$(SOURCES:.cu=.o)
-ARCH=35
+ARCH=70
 
 mkfile_path := $(abspath $(lastword $(MAKEFILE_LIST)))
 current_dir := $(notdir $(patsubst %/,%,$(dir $(mkfile_path))))

util/tracer_nvbit/tracer_tool/traces-processing/post-traces-processing.cpp

@@ -271,6 +271,7 @@ void group_per_block(const char *filepath) {
   // Important... without clear(), cin.eof() may evaluate to true on the second
   // kernel 
   cin.clear();
+  clearerr(stdin);
   while (!cin.eof()) {
     getline(cin, line);